evmrs: add benchmarks

[LorenzSchueler]

This PR adds the benchmarks which contains benchmarks (currently only ffi_overhead) for calling from Rust via FFI into evmrs.

The benchmarks can be used in two ways:

• with criterion to compare different features:
  cargo bench --package benchmarks --profile profiling [--features <feature1>,<feature2>,...]
• as a standalone executable for profiling
  cargo build --package benchmarks --profile profiling [--features <feature1>,<feature2>,...]
  then use ./target/profiling/benchmarks with the profiler of your choice

[LuisPH3]

Unfortunately, it is still not clear why we need so many benchmark functions.
Can you write an example of the use of the main function in benchmarks/src/main.rs
I find difficult to justify different entry points to the program.

[LorenzSchueler]

We need two entry points because with criterion be get statistics about the benchmarks (in particular accurate timing data), but criterion also has some overhead. main.rs executes the same benchmark function(s) but without criterion is therefore better suited for profiling e.g. with perf.

[LorenzSchueler]

You could have multiple files but as long as they all use criterion that does not make too much sense because you can already filter the tests within one test binary. The only thing you would gain is that they are compiled as two separate executables. But they would both depend on benchmarks/src/lib.rs anyway.

[LorenzSchueler]

I renamed the file to interpreter.rs because all tests in here will test the whole interpreter and not just parts of it.

[LuisPH3]

👍